exercises/solutions/movie_classification_unsupervised_incl_recommendations_solution.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Using Embeddings for unsupervised clustering with named clusters (and other fun things)\n",
    "\n",
    "In this notebook, we use a subset of the [Top 10000 Popular Movies Dataset](https://www.kaggle.com/datasets/db55ac3dfd0098a0cf96dd542807f9253a16587ff233e06baef372bccfd09942) to calculate embeddings on movie descriptions and then apply kmeans to find similar clusters. Once we have these clusters, we'll use a prompt to extract the topics from each cluster. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import tiktoken\n",
    "import openai\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from dotenv import load_dotenv\n",
    "from openai.embeddings_utils import cosine_similarity\n",
    "from tenacity import retry, wait_random_exponential, stop_after_attempt\n",
    "\n",
    "# Load environment variables\n",
    "load_dotenv()\n",
    "\n",
    "# Configure OpenAI API\n",
    "openai.api_type = \"azure\"\n",
    "openai.api_version = \"2022-12-01\"\n",
    "openai.api_base = os.getenv('OPENAI_API_BASE')\n",
    "openai.api_key = os.getenv(\"OPENAI_API_KEY\")\n",
    "\n",
    "# Define embedding model and encoding\n",
    "EMBEDDING_MODEL = 'text-embedding-ada-002'\n",
    "COMPLETION_MODEL = 'text-davinci-003'\n",
    "encoding = tiktoken.get_encoding('cl100k_base')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load `movies.csv`:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(500, 12)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>original_language</th>\n",
       "      <th>original_title</th>\n",
       "      <th>popularity</th>\n",
       "      <th>release_date</th>\n",
       "      <th>vote_average</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>genre</th>\n",
       "      <th>overview</th>\n",
       "      <th>revenue</th>\n",
       "      <th>runtime</th>\n",
       "      <th>tagline</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>381284.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Hidden Figures</td>\n",
       "      <td>49.802</td>\n",
       "      <td>2016-12-10</td>\n",
       "      <td>8.1</td>\n",
       "      <td>7310.0</td>\n",
       "      <td>['Drama', 'History']</td>\n",
       "      <td>The untold story of Katherine G. Johnson, Doro...</td>\n",
       "      <td>2.306988e+08</td>\n",
       "      <td>127.0</td>\n",
       "      <td>Meet the women you don't know, behind the miss...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>356334.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Gridlocked</td>\n",
       "      <td>9.801</td>\n",
       "      <td>2016-06-14</td>\n",
       "      <td>5.8</td>\n",
       "      <td>130.0</td>\n",
       "      <td>['Action']</td>\n",
       "      <td>Former SWAT leader David Hendrix and hard-part...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>114.0</td>\n",
       "      <td>Only one way out…</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>475557.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Joker</td>\n",
       "      <td>116.462</td>\n",
       "      <td>2019-10-02</td>\n",
       "      <td>8.2</td>\n",
       "      <td>18970.0</td>\n",
       "      <td>['Crime', 'Thriller', 'Drama']</td>\n",
       "      <td>During the 1980s, a failed stand-up comedian i...</td>\n",
       "      <td>1.074251e+09</td>\n",
       "      <td>122.0</td>\n",
       "      <td>Put on a happy face.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>347847.0</td>\n",
       "      <td>en</td>\n",
       "      <td>The Sand</td>\n",
       "      <td>14.172</td>\n",
       "      <td>2015-08-28</td>\n",
       "      <td>5.1</td>\n",
       "      <td>157.0</td>\n",
       "      <td>['Horror']</td>\n",
       "      <td>Just when you thought it was safe to go back i...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>84.0</td>\n",
       "      <td>This Beach is Killer</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>739542.0</td>\n",
       "      <td>en</td>\n",
       "      <td>America: The Motion Picture</td>\n",
       "      <td>98.542</td>\n",
       "      <td>2021-06-30</td>\n",
       "      <td>5.8</td>\n",
       "      <td>130.0</td>\n",
       "      <td>['Action', 'Comedy', 'History', 'Animation', '...</td>\n",
       "      <td>A chainsaw-wielding George Washington teams wi...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>98.0</td>\n",
       "      <td>This summer they're redrawing history.</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         id original_language               original_title  popularity  \\\n",
       "0  381284.0                en               Hidden Figures      49.802   \n",
       "1  356334.0                en                   Gridlocked       9.801   \n",
       "2  475557.0                en                        Joker     116.462   \n",
       "3  347847.0                en                     The Sand      14.172   \n",
       "4  739542.0                en  America: The Motion Picture      98.542   \n",
       "\n",
       "  release_date  vote_average  vote_count  \\\n",
       "0   2016-12-10           8.1      7310.0   \n",
       "1   2016-06-14           5.8       130.0   \n",
       "2   2019-10-02           8.2     18970.0   \n",
       "3   2015-08-28           5.1       157.0   \n",
       "4   2021-06-30           5.8       130.0   \n",
       "\n",
       "                                               genre  \\\n",
       "0                               ['Drama', 'History']   \n",
       "1                                         ['Action']   \n",
       "2                     ['Crime', 'Thriller', 'Drama']   \n",
       "3                                         ['Horror']   \n",
       "4  ['Action', 'Comedy', 'History', 'Animation', '...   \n",
       "\n",
       "                                            overview       revenue  runtime  \\\n",
       "0  The untold story of Katherine G. Johnson, Doro...  2.306988e+08    127.0   \n",
       "1  Former SWAT leader David Hendrix and hard-part...  0.000000e+00    114.0   \n",
       "2  During the 1980s, a failed stand-up comedian i...  1.074251e+09    122.0   \n",
       "3  Just when you thought it was safe to go back i...  0.000000e+00     84.0   \n",
       "4  A chainsaw-wielding George Washington teams wi...  0.000000e+00     98.0   \n",
       "\n",
       "                                             tagline  \n",
       "0  Meet the women you don't know, behind the miss...  \n",
       "1                                  Only one way out…  \n",
       "2                               Put on a happy face.  \n",
       "3                               This Beach is Killer  \n",
       "4             This summer they're redrawing history.  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv('../../data/movies/movies.csv')\n",
    "print(df.shape)\n",
    "df.head()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, let's create a new column and calculate how many tokens each embedding would cost. This allows us to get an estimate how much we'd pay to create embeddings on the whole dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test would cost $0.0117564 for embeddings\n"
     ]
    }
   ],
   "source": [
    "# add a new column to the dataframe where you put the token count of the review\n",
    "df = df.assign(token_count=df['overview'].apply(lambda x: len(encoding.encode(x))))\n",
    "\n",
    "# print the first 5 rows of the dataframe, then also the total number of tokens\n",
    "total_tokens = df['token_count'].sum()\n",
    "\n",
    "cost_for_embeddings = total_tokens / 1000 * 0.0004\n",
    "print(f\"Test would cost ${cost_for_embeddings} for embeddings\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's define our embedding method. Please note the use of tenacity for having an automated retry mechanism, in case we hit the TPS limits of Azure OpenAI Service."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "@retry(wait=wait_random_exponential(min=1, max=20), stop=stop_after_attempt(10))\n",
    "def get_embedding(text) -> list[float]:\n",
    "    text = text.replace(\"\\n\", \" \")\n",
    "    return openai.Embedding.create(input=text, engine=EMBEDDING_MODEL)[\"data\"][0][\"embedding\"]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's creating the embeddings:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>original_language</th>\n",
       "      <th>original_title</th>\n",
       "      <th>popularity</th>\n",
       "      <th>release_date</th>\n",
       "      <th>vote_average</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>genre</th>\n",
       "      <th>overview</th>\n",
       "      <th>revenue</th>\n",
       "      <th>runtime</th>\n",
       "      <th>tagline</th>\n",
       "      <th>token_count</th>\n",
       "      <th>embedding</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>381284.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Hidden Figures</td>\n",
       "      <td>49.802</td>\n",
       "      <td>2016-12-10</td>\n",
       "      <td>8.1</td>\n",
       "      <td>7310.0</td>\n",
       "      <td>['Drama', 'History']</td>\n",
       "      <td>The untold story of Katherine G. Johnson, Doro...</td>\n",
       "      <td>2.306988e+08</td>\n",
       "      <td>127.0</td>\n",
       "      <td>Meet the women you don't know, behind the miss...</td>\n",
       "      <td>62</td>\n",
       "      <td>[-0.016777517274022102, -0.013854648917913437,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>356334.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Gridlocked</td>\n",
       "      <td>9.801</td>\n",
       "      <td>2016-06-14</td>\n",
       "      <td>5.8</td>\n",
       "      <td>130.0</td>\n",
       "      <td>['Action']</td>\n",
       "      <td>Former SWAT leader David Hendrix and hard-part...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>114.0</td>\n",
       "      <td>Only one way out…</td>\n",
       "      <td>35</td>\n",
       "      <td>[-0.006641489453613758, -0.022688400000333786,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>475557.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Joker</td>\n",
       "      <td>116.462</td>\n",
       "      <td>2019-10-02</td>\n",
       "      <td>8.2</td>\n",
       "      <td>18970.0</td>\n",
       "      <td>['Crime', 'Thriller', 'Drama']</td>\n",
       "      <td>During the 1980s, a failed stand-up comedian i...</td>\n",
       "      <td>1.074251e+09</td>\n",
       "      <td>122.0</td>\n",
       "      <td>Put on a happy face.</td>\n",
       "      <td>36</td>\n",
       "      <td>[-0.011327777057886124, -0.028586726635694504,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>347847.0</td>\n",
       "      <td>en</td>\n",
       "      <td>The Sand</td>\n",
       "      <td>14.172</td>\n",
       "      <td>2015-08-28</td>\n",
       "      <td>5.1</td>\n",
       "      <td>157.0</td>\n",
       "      <td>['Horror']</td>\n",
       "      <td>Just when you thought it was safe to go back i...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>84.0</td>\n",
       "      <td>This Beach is Killer</td>\n",
       "      <td>124</td>\n",
       "      <td>[-0.0012890542857348919, -0.03704281151294708,...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>739542.0</td>\n",
       "      <td>en</td>\n",
       "      <td>America: The Motion Picture</td>\n",
       "      <td>98.542</td>\n",
       "      <td>2021-06-30</td>\n",
       "      <td>5.8</td>\n",
       "      <td>130.0</td>\n",
       "      <td>['Action', 'Comedy', 'History', 'Animation', '...</td>\n",
       "      <td>A chainsaw-wielding George Washington teams wi...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>98.0</td>\n",
       "      <td>This summer they're redrawing history.</td>\n",
       "      <td>32</td>\n",
       "      <td>[-0.006386359687894583, -0.02050323784351349, ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         id original_language               original_title  popularity  \\\n",
       "0  381284.0                en               Hidden Figures      49.802   \n",
       "1  356334.0                en                   Gridlocked       9.801   \n",
       "2  475557.0                en                        Joker     116.462   \n",
       "3  347847.0                en                     The Sand      14.172   \n",
       "4  739542.0                en  America: The Motion Picture      98.542   \n",
       "\n",
       "  release_date  vote_average  vote_count  \\\n",
       "0   2016-12-10           8.1      7310.0   \n",
       "1   2016-06-14           5.8       130.0   \n",
       "2   2019-10-02           8.2     18970.0   \n",
       "3   2015-08-28           5.1       157.0   \n",
       "4   2021-06-30           5.8       130.0   \n",
       "\n",
       "                                               genre  \\\n",
       "0                               ['Drama', 'History']   \n",
       "1                                         ['Action']   \n",
       "2                     ['Crime', 'Thriller', 'Drama']   \n",
       "3                                         ['Horror']   \n",
       "4  ['Action', 'Comedy', 'History', 'Animation', '...   \n",
       "\n",
       "                                            overview       revenue  runtime  \\\n",
       "0  The untold story of Katherine G. Johnson, Doro...  2.306988e+08    127.0   \n",
       "1  Former SWAT leader David Hendrix and hard-part...  0.000000e+00    114.0   \n",
       "2  During the 1980s, a failed stand-up comedian i...  1.074251e+09    122.0   \n",
       "3  Just when you thought it was safe to go back i...  0.000000e+00     84.0   \n",
       "4  A chainsaw-wielding George Washington teams wi...  0.000000e+00     98.0   \n",
       "\n",
       "                                             tagline  token_count  \\\n",
       "0  Meet the women you don't know, behind the miss...           62   \n",
       "1                                  Only one way out…           35   \n",
       "2                               Put on a happy face.           36   \n",
       "3                               This Beach is Killer          124   \n",
       "4             This summer they're redrawing history.           32   \n",
       "\n",
       "                                           embedding  \n",
       "0  [-0.016777517274022102, -0.013854648917913437,...  \n",
       "1  [-0.006641489453613758, -0.022688400000333786,...  \n",
       "2  [-0.011327777057886124, -0.028586726635694504,...  \n",
       "3  [-0.0012890542857348919, -0.03704281151294708,...  \n",
       "4  [-0.006386359687894583, -0.02050323784351349, ...  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = df.assign(embedding=df['overview'].apply(lambda x: get_embedding(x)))\n",
    "df.head()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, let's create clusters on the embeddings using KMeans. In this case, we'll go for 5 clusters, knowing that this might be wrong."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\clsieble\\Miniconda3\\envs\\openai-qna-env\\lib\\site-packages\\sklearn\\cluster\\_kmeans.py:870: FutureWarning: The default value of `n_init` will change from 10 to 'auto' in 1.4. Set the value of `n_init` explicitly to suppress the warning\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>original_language</th>\n",
       "      <th>original_title</th>\n",
       "      <th>popularity</th>\n",
       "      <th>release_date</th>\n",
       "      <th>vote_average</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>genre</th>\n",
       "      <th>overview</th>\n",
       "      <th>revenue</th>\n",
       "      <th>runtime</th>\n",
       "      <th>tagline</th>\n",
       "      <th>token_count</th>\n",
       "      <th>embedding</th>\n",
       "      <th>cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>381284.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Hidden Figures</td>\n",
       "      <td>49.802</td>\n",
       "      <td>2016-12-10</td>\n",
       "      <td>8.1</td>\n",
       "      <td>7310.0</td>\n",
       "      <td>['Drama', 'History']</td>\n",
       "      <td>The untold story of Katherine G. Johnson, Doro...</td>\n",
       "      <td>2.306988e+08</td>\n",
       "      <td>127.0</td>\n",
       "      <td>Meet the women you don't know, behind the miss...</td>\n",
       "      <td>62</td>\n",
       "      <td>[-0.016777517274022102, -0.013854648917913437,...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>356334.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Gridlocked</td>\n",
       "      <td>9.801</td>\n",
       "      <td>2016-06-14</td>\n",
       "      <td>5.8</td>\n",
       "      <td>130.0</td>\n",
       "      <td>['Action']</td>\n",
       "      <td>Former SWAT leader David Hendrix and hard-part...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>114.0</td>\n",
       "      <td>Only one way out…</td>\n",
       "      <td>35</td>\n",
       "      <td>[-0.006641489453613758, -0.022688400000333786,...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>475557.0</td>\n",
       "      <td>en</td>\n",
       "      <td>Joker</td>\n",
       "      <td>116.462</td>\n",
       "      <td>2019-10-02</td>\n",
       "      <td>8.2</td>\n",
       "      <td>18970.0</td>\n",
       "      <td>['Crime', 'Thriller', 'Drama']</td>\n",
       "      <td>During the 1980s, a failed stand-up comedian i...</td>\n",
       "      <td>1.074251e+09</td>\n",
       "      <td>122.0</td>\n",
       "      <td>Put on a happy face.</td>\n",
       "      <td>36</td>\n",
       "      <td>[-0.011327777057886124, -0.028586726635694504,...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>347847.0</td>\n",
       "      <td>en</td>\n",
       "      <td>The Sand</td>\n",
       "      <td>14.172</td>\n",
       "      <td>2015-08-28</td>\n",
       "      <td>5.1</td>\n",
       "      <td>157.0</td>\n",
       "      <td>['Horror']</td>\n",
       "      <td>Just when you thought it was safe to go back i...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>84.0</td>\n",
       "      <td>This Beach is Killer</td>\n",
       "      <td>124</td>\n",
       "      <td>[-0.0012890542857348919, -0.03704281151294708,...</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>739542.0</td>\n",
       "      <td>en</td>\n",
       "      <td>America: The Motion Picture</td>\n",
       "      <td>98.542</td>\n",
       "      <td>2021-06-30</td>\n",
       "      <td>5.8</td>\n",
       "      <td>130.0</td>\n",
       "      <td>['Action', 'Comedy', 'History', 'Animation', '...</td>\n",
       "      <td>A chainsaw-wielding George Washington teams wi...</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>98.0</td>\n",
       "      <td>This summer they're redrawing history.</td>\n",
       "      <td>32</td>\n",
       "      <td>[-0.006386359687894583, -0.02050323784351349, ...</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         id original_language               original_title  popularity  \\\n",
       "0  381284.0                en               Hidden Figures      49.802   \n",
       "1  356334.0                en                   Gridlocked       9.801   \n",
       "2  475557.0                en                        Joker     116.462   \n",
       "3  347847.0                en                     The Sand      14.172   \n",
       "4  739542.0                en  America: The Motion Picture      98.542   \n",
       "\n",
       "  release_date  vote_average  vote_count  \\\n",
       "0   2016-12-10           8.1      7310.0   \n",
       "1   2016-06-14           5.8       130.0   \n",
       "2   2019-10-02           8.2     18970.0   \n",
       "3   2015-08-28           5.1       157.0   \n",
       "4   2021-06-30           5.8       130.0   \n",
       "\n",
       "                                               genre  \\\n",
       "0                               ['Drama', 'History']   \n",
       "1                                         ['Action']   \n",
       "2                     ['Crime', 'Thriller', 'Drama']   \n",
       "3                                         ['Horror']   \n",
       "4  ['Action', 'Comedy', 'History', 'Animation', '...   \n",
       "\n",
       "                                            overview       revenue  runtime  \\\n",
       "0  The untold story of Katherine G. Johnson, Doro...  2.306988e+08    127.0   \n",
       "1  Former SWAT leader David Hendrix and hard-part...  0.000000e+00    114.0   \n",
       "2  During the 1980s, a failed stand-up comedian i...  1.074251e+09    122.0   \n",
       "3  Just when you thought it was safe to go back i...  0.000000e+00     84.0   \n",
       "4  A chainsaw-wielding George Washington teams wi...  0.000000e+00     98.0   \n",
       "\n",
       "                                             tagline  token_count  \\\n",
       "0  Meet the women you don't know, behind the miss...           62   \n",
       "1                                  Only one way out…           35   \n",
       "2                               Put on a happy face.           36   \n",
       "3                               This Beach is Killer          124   \n",
       "4             This summer they're redrawing history.           32   \n",
       "\n",
       "                                           embedding  cluster  \n",
       "0  [-0.016777517274022102, -0.013854648917913437,...        0  \n",
       "1  [-0.006641489453613758, -0.022688400000333786,...        3  \n",
       "2  [-0.011327777057886124, -0.028586726635694504,...        3  \n",
       "3  [-0.0012890542857348919, -0.03704281151294708,...        2  \n",
       "4  [-0.006386359687894583, -0.02050323784351349, ...        3  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# train k-means on df embeddings\n",
    "from sklearn.cluster import KMeans\n",
    "\n",
    "n_clusters = 5\n",
    "kmeans = KMeans(n_clusters=n_clusters, init=\"k-means++\", random_state=42)\n",
    "kmeans.fit(df['embedding'].to_list())\n",
    "df = df.assign(cluster=kmeans.labels_)\n",
    "df.head()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we have a cluster per row, let's use t-SNE to project our embeddings into 2d space and visualize the clusters."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(500, 1536)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\clsieble\\AppData\\Local\\Temp\\ipykernel_30768\\2572688562.py:20: RuntimeWarning: Mean of empty slice.\n",
      "  avg_x = xs.mean()\n",
      "c:\\Users\\clsieble\\Miniconda3\\envs\\openai-qna-env\\lib\\site-packages\\numpy\\core\\_methods.py:192: RuntimeWarning: invalid value encountered in divide\n",
      "  ret = ret.dtype.type(ret / rcount)\n",
      "C:\\Users\\clsieble\\AppData\\Local\\Temp\\ipykernel_30768\\2572688562.py:21: RuntimeWarning: Mean of empty slice.\n",
      "  avg_y = ys.mean()\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.manifold import TSNE\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "tsne = TSNE(\n",
    "    n_components=2, perplexity=50, random_state=42, init=\"random\", learning_rate=200\n",
    ")\n",
    "\n",
    "matrix = np.vstack(df.embedding.values)\n",
    "print(matrix.shape)\n",
    "vis_dims2 = tsne.fit_transform(matrix)\n",
    "\n",
    "x = [x for x, y in vis_dims2]\n",
    "y = [y for x, y in vis_dims2]\n",
    "\n",
    "for category, color in enumerate([\"purple\", \"green\", \"red\", \"blue\",\"yellow\", 'black', 'orange', 'brown', 'pink', 'grey']):\n",
    "    xs = np.array(x)[df.cluster == category]\n",
    "    ys = np.array(y)[df.cluster == category]\n",
    "    plt.scatter(xs, ys, color=color, alpha=0.3)\n",
    "\n",
    "    avg_x = xs.mean()\n",
    "    avg_y = ys.mean()\n",
    "\n",
    "    plt.scatter(avg_x, avg_y, marker=\"x\", color=color, s=100)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ugh ok, that does not look great, but was somehow expected. We have all kinds of movies and with only 5 clusters, this might not be ideal. However, if you look closely, you can make up some rough shape that resembles a cluster. Also, movies might fall into two or more categories, so it kind of makes sense.\n",
    "\n",
    "Lastly, let's take a few examples from each cluster, send them to OpenAI and get the common theses extracted:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster 0 topics: Brave, Revolution, Rescue\n",
      "Movies from cluster 0: Le Grand Bleu, Norm of the North, Arctic Dogs, Man on Wire, Willy Wonka & the Chocolate Factory, The War Below, A Stork's Journey, The Monuments Men, The Devil's Brigade, Hidden Figures, The Nativity Story, East of Eden, Black Sea, Ben-Hur, La Bamba, Jetsons: The Movie, Borg vs McEnroe, Australia, 崖の上のポニョ, Animal Farm, The Aristocats, Poetic Justice, Lady and the Tramp, The Road to El Dorado, Броненосец Потёмкин\n",
      "================\n",
      "Cluster 1 topics: Rebellion, Teenagers, Love.\n",
      "Movies from cluster 1: Tschick, Always Be My Maybe, Jane Eyre, Coco avant Chanel, Something the Lord Made, My Best Friend's Girl, Yesterday, The War of the Roses, The Full Monty, Orange County, Pitch Perfect, Niñas Mal, Behaving Badly, Big Eyes, John Tucker Must Die, Fame, She's the Man, Rabbit Hole, The Roommate, The Baby-Sitters Club, Saint Maud, What Lies Below, Katy Perry: Part of Me, Me Before You, Sexology\n",
      "================\n",
      "Cluster 2 topics: Zombies, Survivors, Escape\n",
      "Movies from cluster 2: Tremors 3: Back to Perfection, Leatherface, Ooops! Noah is Gone..., Good Luck Charlie, It's Christmas!, 巨乳ドラゴン 温泉ゾンビVSストリッパー5, Rampage, Friday the 13th Part VII: The New Blood, Mimic, Pandorum, Paranormal Activity: The Ghost Dimension, Settlers, The Sand, Messengers 2: The Scarecrow, Land of the Dead, Shaun of the Dead, Big Driver, Jurassic Park, Dinosaur Hotel, Let Me In, Summer '03, The Diabolical, Attack the Block, The Addams Family, Cargo, Kristy\n",
      "================\n",
      "Cluster 3 topics: Action, crime, suspense.\n",
      "Movies from cluster 3: Blunt Force Trauma, Furious 7, The King's Man, The Nice Guys, Batman & Robin, America: The Motion Picture, Airplane II: The Sequel, Super Troopers 2, Presumed Innocent, WASABI, Midnight in the Switchgrass, Suicide Squad, Clerks, In Hell, My Spy, Under Suspicion, To Catch a Thief, Mission: Impossible - Rogue Nation, Open Grave, The Raid 2: Berandal, El Infierno, Asher, The Wrestler, The Fanatic, Fire Down Below\n",
      "================\n",
      "Cluster 4 topics: Fantasy, adventure, science\n",
      "Movies from cluster 4: The Huntsman: Winter's War, The Core, メトロポリス, The Avengers, Rumble, Pirates of the Caribbean: On Stranger Tides, Cats, ドラゴンボールＺ 危険なふたり！超戦士はねむれない, Troy, The Rocketeer, Atlantis: Milo's Return, The Scorpion King, Beauty and the Beast, 僕のヒーローアカデミア THE MOVIE ～2人の英雄～, The Swan Princess, Bleach, Treasure Island, Enemy Mine, Frozen, The Princess and the Frog, Quest for Camelot, Tinker Bell, Dungeons & Dragons, Hotel Transylvania 3: Summer Vacation, Transformers: The Last Knight\n",
      "================\n"
     ]
    }
   ],
   "source": [
    "# take 10 movies from each cluster and write a prompt that asks what these have in common\n",
    "# ideally you would use more movies than 10, but this is just a demo\n",
    "for i in range(n_clusters):\n",
    "    reviews = df[df['cluster'] == i]['overview'].sample(10)\n",
    "    reviews = \"\\n\".join(reviews.values.tolist())\n",
    "    \n",
    "    prompt = f\"Here are 10 movie descriptions:\\n{reviews}Write 3 words what these have in common?\"\n",
    "    #print(prompt)\n",
    "    response = openai.Completion.create(engine=COMPLETION_MODEL, prompt=prompt, temperature=0.7, max_tokens=100, top_p=1, frequency_penalty=0, presence_penalty=0, stop=None)['choices'][0]['text'].strip()\n",
    "    print(f\"Cluster {i} topics: {response}\")\n",
    "    movies = df[df['cluster'] == i]['original_title'].sample(25)\n",
    "    print(f\"Movies from cluster {i}: {', '.join(movies.values.tolist())}\")\n",
    "    print(\"================\")    "
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Doesn't look too bad. But again, 10 samples for each class might not be enough, given the low number of clusters. Anyway, looking at the movie titles, some of the topics actually make fairly ok sense."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Bonus: Using the embeddings to build a simple recommendation system"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Another thing we can do is use the embeddings for building a very simple recommendation system. So let's try it:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>original_title</th>\n",
       "      <th>similarity</th>\n",
       "      <th>overview</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>480</th>\n",
       "      <td>Frozen</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>Young princess Anna of Arendelle dreams about ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>361</th>\n",
       "      <td>The Huntsman: Winter's War</td>\n",
       "      <td>0.862036</td>\n",
       "      <td>As two evil sisters prepare to conquer the lan...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>468</th>\n",
       "      <td>The Swan Princess</td>\n",
       "      <td>0.857933</td>\n",
       "      <td>The beautiful princess Odette is transformed i...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>161</th>\n",
       "      <td>Beauty and the Beast</td>\n",
       "      <td>0.849219</td>\n",
       "      <td>A live-action adaptation of Disney's version o...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>173</th>\n",
       "      <td>Cinderella</td>\n",
       "      <td>0.842784</td>\n",
       "      <td>When her father unexpectedly passes away, youn...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>286</th>\n",
       "      <td>Wonder Woman</td>\n",
       "      <td>0.842748</td>\n",
       "      <td>An Amazon princess comes to the world of Man i...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 original_title  similarity  \\\n",
       "480                      Frozen    1.000000   \n",
       "361  The Huntsman: Winter's War    0.862036   \n",
       "468           The Swan Princess    0.857933   \n",
       "161        Beauty and the Beast    0.849219   \n",
       "173                  Cinderella    0.842784   \n",
       "286                Wonder Woman    0.842748   \n",
       "\n",
       "                                              overview  \n",
       "480  Young princess Anna of Arendelle dreams about ...  \n",
       "361  As two evil sisters prepare to conquer the lan...  \n",
       "468  The beautiful princess Odette is transformed i...  \n",
       "161  A live-action adaptation of Disney's version o...  \n",
       "173  When her father unexpectedly passes away, youn...  \n",
       "286  An Amazon princess comes to the world of Man i...  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Let's pick a movie that exists in df, keeping in mind we only have 500 movies in it!\n",
    "movie = \"Frozen\"\n",
    "\n",
    "# get embedding for movie\n",
    "e = df[df['original_title'] == movie]['embedding'].values[0]\n",
    "\n",
    "# get cosine similarity between movie and all other movies and sort ascending\n",
    "similarities = df['embedding'].apply(lambda x: cosine_similarity(x, e))\n",
    "\n",
    "# get most similar movies\n",
    "movies = df.assign(similarity=similarities).sort_values(by='similarity', ascending=False)[['original_title', 'similarity', 'overview']]\n",
    "movies[0:6]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Besides the last one, this actually does not look too bad...probably would have been useful if we added movie categories and age ratings to our recommendations... :)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "openai-qna-env",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.9"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "c4ee1bbf3137c7ea9420c4fd488a55642063e5739fe2a7286130d9ba47405b69"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}